Principles

[moodymudskipper]

• no effort is made to avoid confusion resulting from habits or special syntax used in other packages, we start from scratch and challenge everything, but of course keep good ideas.
• [x] What worked (and made sense) on a standard data.frame should work all the same on tb objects : https://github.com/moodymudskipper/tb/issues/2
• [x] no assignment by reference : https://github.com/moodymudskipper/tb/issues/3
• [x] .i is used only for subsetting (standard subsetting and semi joins) : https://github.com/moodymudskipper/tb/issues/4
• [x] .j is used only for selection and transmuting (like SELECT in SQL)
  :https://github.com/moodymudskipper/tb/issues/5 https://github.com/moodymudskipper/tb/issues/8
• [x] .by is used only for aggregation (resulting in one row per set of grouping variables) : https://github.com/moodymudskipper/tb/issues/6
• [x] ... is used for mutating and summarizing (if .by isn't NULL) : https://github.com/moodymudskipper/tb/issues/7 https://github.com/moodymudskipper/tb/issues/8
• [x] A two sided formula notation is used for mutation by group or subsetting by group https://github.com/moodymudskipper/tb/issues/9
• [x] := is used as an extension of = which behaves the same whenever = works https://github.com/moodymudskipper/tb/issues/10
• [x] .() will work as in bquote(), not as in data.table, for easy substitution https://github.com/moodymudskipper/tb/issues/11
• [x] .data can be used anywhere to refer to source table and .subset to refer to what is called .SD in data.table, https://github.com/moodymudskipper/tb/issues/12
• [ ] All possible data transformations characterized by a main table input should be supported, including pivoting (spread/gather/melt/cast), nesting (nest/unnest/pack/unpack/chop/unchop) and joins, which share the property that they have one main table input. https://github.com/moodymudskipper/tb/issues/13